African Journal of Aquatic Science

ISSN: 1608-5914 (Print) 1727-9364 (Online) Journal homepage: https://www.tandfonline.com/loi/taas20

Using action cameras to estimate the abundance and habitat use of threatened fish in clear headwater streams

B Hannweg, SM Marr, LE Bloy & OLF Weyl

To cite this article: B Hannweg, SM Marr, LE Bloy & OLF Weyl (2020): Using action cameras to estimate the abundance and habitat use of threatened fish in clear headwater streams, African Journal of Aquatic Science, DOI: 10.2989/16085914.2019.1701404 To link to this article: https://doi.org/10.2989/16085914.2019.1701404

View supplementary material

Published online: 12 May 2020.

Submit your article to this journal

Article views: 6

View related articles

View Crossmark data

Full Terms & Conditions of access and use can be found at https://www.tandfonline.com/action/journalInformation?journalCode=taas20 African Journal of Aquatic Science 2020, 45(1): xxx–xxx Copyright © NISC (Pty) Ltd Printed in South Africa — All rights reserved AFRICAN JOURNAL OF AQUATIC SCIENCE This is the final version of the article that is published ISSN 1608-5914 EISSN 1727-9364 ahead of the print and online issue https://doi.org/10.2989/16085914.2019.1701404

Short Note

Using action cameras to estimate the abundance and habitat use of threatened fish in clear headwater streams

B Hannweg1,2, SM Marr2,3* , LE Bloy1,2,3 and OLF Weyl1,2,3

1 Department of Ichthyology and Fisheries Science, Rhodes University, Grahamstown, South Africa 2 DSI/NRF Research Chair in Inland Fisheries and Freshwater Ecology, South African Institute for Aquatic Biodiversity, Grahamstown, South Africa 3 Centre for Invasion Biology, South African Institute for Aquatic Biodiversity, Grahamstown, South Africa *Correspondence: [email protected]

Snorkel and electrofishing surveys are the traditional baseline methods for fish surveys in clear headwater streams. However, action cameras provide a non-harmful alternative to monitor freshwater fish populations to develop informed conservation management initiatives. In this paper, estimates from photographs and videos from action cameras are compared with snorkel survey estimates of the density of a threatened endemic minnow in a headwater stream, Eastern Cape, South Africa. Photograph-based relative abundances of fish summed over five microhabitats in each pool returned equivalent results to snorkel surveys, whereas the equivalent video-based abundance estimates were approximately 50% greater than the snorkel estimates. Therefore, photograph-derived estimates could be used as an alternative to snorkel surveys for fish population monitoring and habitat use studies in clear headwater streams.

Keywords: freshwater fish, MaxN, monitoring, photographs, snorkel surveys, underwater video analysis

Supplementary material: available online at https://doi.org/10.2989/16085914.2019.1701404

Freshwater fish are currently more impacted by human physiological stress, spinal injuries and even internal activities than are other faunal groups (Duncan and haemorrhaging (Snyder 2003). As a result, visual methods, Lockwood 2001; Limburg et al. 2011). Threats, which such as snorkel surveys (Dolloff et al. 1996), and more include habitat degradation/fragmentation, water pollution, recently action cameras, are the preferred survey methods over-abstraction, and the introduction of non-native fish for fish population assessments (Ebner and Morgan 2013), species (Cowx 2002), are compounded for species isolated and studies of ecology, behaviour and habitat utilisation within headwater streams or with restricted geographical (Ebner et al. 2014), particularly for imperilled fish (Ebner et distributions (Ellender et al. 2017). The reason is that al. 2009) these small isolated populations are more vulnerable to In South Africa’s Cape Fold Ecoregion (CFE), sensu (Abell extirpation and local extinctions than more widespread et al. 2008), the imperilled conservation status of many species. Knowledge of the distribution, behaviour and headwater stream fish (Ellender et al. 2017) requires lower abundance of fish, particularly those of conservation risk methods of assessing their abundance, distribution concern, is important for evaluations of their conservation and habitat use (Ellender et al. 2012). Althoug underwater status (Balian et al. 2008; Darwall et al. 2011) and to video analysis (UWVA) provides a useful alternative to develop and direct conservation initiatives (Darwall et al. snorkel surveys and electrofishing, abundance estimates 2008; Darwall et al. 2011). are limited by the camera’s field of view and are, therefore, Fish populations in headwater streams have traditionally generally reported as relative abundance estimates (Ellender been monitored using electrofishing (Hickey and Closs et al. 2012). In addition, post-survey analysis of videos can 2006), seine netting (Anderson et al. 1995; Jordan et be prohibitively time consuming (Cappo et al. 2003; Ebner al. 2008), minnow traps (Kadye and Booth 2014) and et al. 2009). In an attempt to improve UWVA methods, we snorkel surveys (Jordan et al. 2008; Ebner et al. 2009). used headwater stream populations of Eastern Cape redfin Electrofishing, seine netting and trapping involve physically afer (Peters, 1864) in the Swartkops River handling fish and could result in post-release mortality as a model system to: (1) compare different non-harmful by increasing the risk of infection through the removal of methods of estimating freshwater fish abundance; (2) assess their protective mucilaginous layer (Brydges et al. 2009), the utility of a 5–camera array to derive absolute abundance whereas electrofishing could result in bleeding at the gills, estimates; and (3) use UWVA to assess habitat use of

African Journal of Aquatic Science is co-published by NISC (Pty) Ltd and Informa UK Limited (trading as Taylor & Francis Group) Published online 12 May 2020 2 Hannweg, Marr, Bloy and Weyl fish in small headwater stream habitats. For the current The absolute abundance of P. afer in each pool was study, snorkel surveys, traditionally used to estimate fish estimated by snorkel surveys using the two-pass method abundance in clear headwater streams were used as the described by (Ellender et al. 2018). After the snorkel survey, standard method to evaluate the performance of UWVA. five GoPro Hero 3+ (GoPro Inc., USA) action cameras, in The current study was conducted during February 2016 waterproof housings mounted on tripods, were deployed, in the Groendal Wilderness Area over ten pools in two each covering one of five microhabitats in each pool: perennial headwater tributaries of the Swartkops River, inflow, outflow, deep middle, woody debris and submerged Eastern Cape, South Africa (Table 1). The Eastern Cape root-wads of riparian ferns. Camera placement ensured redfin, a small endemic cyprinid restricted to the headwaters non-overlapping fields of vision. Following recommendations of the Baakens, Swartkops and Sundays rivers in the by Ellender et al. (2012), cameras were set to film for 18 Eastern Cape (Chakona and Skelton 2017) is classified min (including a 3-min acclimation period) at 60 fps (frames as Endangered in the 2017 IUCN-red-list (Chakona et al. per second) and 720 p (progressive display format), while 2017). The sites were chosen as the relationship between concurrently taking 12-megapixel resolution photographs fish relative abundance estimates from UWVA and absolute every five seconds. abundance estimates from electrofishing had previously been Videos were viewed using Windows Media Player demonstrated there for this species (Ellender et al. 2012). (Microsoft Inc.), photographs using Windows Photo Sites, comprising of different pool habitats, were selected Viewer (Microsoft Inc.), and abundance estimated using on the basis of the presence of the target species P. afer, MaxN (sensu Cappo et al. 2003). The first three minutes suitable water clarity for snorkel surveys (>5 m visibility) of the videos was discarded as acclimation time, and the and UWVA, and being deep enough to snorkel (>0.3 m remaining 15-min sample time divided into 30-s intervals deep); see Ellender et al. (2018). Water physico-chemical for which the MaxN was estimated individually. The parameters (pH, temperature, electrical conductivity and MaxN for the deployment was the maximum number of turbidity) were measured using a HANNA HI98129 combo P. afer individuals viewed in any of the 30 s segments. probe and a HANNA HI 98703 turbidity meter (HANNA Similarly, photographs were viewed in batches of six, Instruments, Inc., United States of America.) Three excluding those from the first three minutes, and the MaxN measurements of each parameter were recorded in each determined, as described for videos. pool and averaged for each site. Physical and environmental To account for differences in pool size, absolute parameters for each site at the time of study are presented in abundances of P. afer were standardised to density Table 1. Flow in the system is episodic (Ellender et al. 2011) estimates by dividing the abundance by pool surface area, and, during the current study, flow was negligible. calculated as the product of the length of the pool and

Table 1: Summary of habitat measurements, location and physico-chemical parameters (± standard deviation) for ten pools within two headwater tributaries of the Swartkops River, Eastern Cape, South Africa

Mean Water Length Area Conductivity Turbidity Site Pool Location width temperature pH (m) (m2) (µS cm−1) (NTU) (m) (°C) P01 Blindekloof 1 33°41′56.81′′ S, 7.4 2.08 15.4 19.9 ± 0.1 230 ± 1 0.65 ± 0.36 4.84 ± 0.68 25°18′14.37′′ E P02 Blindekloof 2 33°41′39.37′′ S, 6.8 2.46 16.7 19.8 ± 0.1 233 ± 4 0.22 ± 0.08 5.18 ± 0.17 25°18′35.96′′ E P03 Blindekloof 3 33°41′25.07′′ S, 8.9 4.09 36.4 19.4 ± 0.1 236 ± 3 0.25 ± 0.02 4.93 ± 0.09 25°18′29.42′′ E P04 Fernkloof 1 33°43′27.45′′ S, 10.0 3.05 30.5 18.0 ± 0.1 233 ± 1 0.67 ± 0.43 5.15 ± 0.02 25°17′03.33′′ E P05 Fernkloof 2 33°43′27.26′′ S, 10.0 3.52 35.2 18.2 ± 0.1 205 ± 3 0.53 ± 0.09 5.12 ± 0.09 25°17′03.72′′ E P06 Fernkloof 3 33°43′26.65′′ S, 17.0 3.02 51.4 19.8 ± 0.1 336 ± 3 0.23 ± 0.02 6.23 ± 0.08 25°17′05.28′′ E P07 Fernkloof 4 33°43′25.73′′ S, 6.6 4.65 30.8 18.1 ± 0.1 346 ± 3 0.17 ± 0.02 5.54 ± 0.06 25°17′07.12′′ E P08 Fernkloof 5 33°43′24.64′′ S, 13.0 2.55 33.2 18.3 ± 0.1 344 ± 3 0.20 ± 0.03 5.56 ± 0.08 25°17′07.80′′ E P09 Fernkloof 6 33°43′21.29′′ S, 9.4 3.27 30.7 18.4 ± 0.1 338 ± 4 0.47 ± 0.22 6.09 ± 0.23 25°17′13.38′′ E P10 Fernkloof 7 33°43′18.88′′ S, 5.4 3.05 16.5 20.8 ± 0.1 241 ± 2 0.34 ± 0.04 5.55 ± 0.04 25°17′15.17′′ E African Journal of Aquatic Science 2020, 45(1): xxx–xxx 3 the average of five equally spaced width measurements. 3-min acclimation period, the MaxN values were summed Comparisons between snorkel, video and photograph for the cameras in each pool, for each 30 s interval, over derived estimates of P. afer abundance were based on the sampling period. The Shapiro-Wilk test was used to four datasets: (1) snorkel survey P. afer density estimates determine whether the dataset from each site was normally (n = 10 sites); (2) video-derived P. afer density estimates distributed. The sample period was then divided into two from a single camera set in the deepest section of the groups, the first 3 min following the acclimation period and pool (the standard camera placement in stream surveys the remaining 12 min of the recording. The hypotheses (Ellender et al. 2012; Weyl et al. 2013) (n = 10 sites); that the means/medians of these two groups originated (3) video- and photograph-derived total P. afer density from the same distributions were evaluated independently estimates (summation of the P. afer density estimates over for each pool using either a Mann–Whitney U-test (for the five cameras in each pool; n = 10 sites for both videos non-normal data) or a t-test (for normal data). and photographs); and (4) video- and photograph-derived The MaxN abundance of P. afer in each habitat were P. afer MaxN estimates from each camera in each pool not normally distributed (Shapiro–Wilk; p < 0.001), but (n = 50 camera deployments for both). ANOVAs were used, because of the large number of A Pearson correlation was used to assess the strength points in the datasets (Van Hecke 2012); 750 abundance of the correlation between the respective density estimates records for each habitat in each pool. A two-way ANOVA and a paired t-test was performed to determine whether (pool and habitat) was performed and found no significant the paired density estimates generated by the respective results for pools, but significant results for habitat type and survey techniques originated from the same datasets. the pool-habitat interaction, Therefore, a one-way ANOVA Linear regression was then used to provide a visualisation of habitat-type was performed to evaluate differences in of the relationship between the paired density estimates abundances of P. afer between habitat types. Post hoc generated by the respective survey techniques, with Tukey HSD tests were conducted to identify the habitat- the correlation of the regression. This was to signify how type pairs contributing to the significant results for the well the regression explained the relationship between one-way ANOVA. All data analyses were conducted using the respective estimates and the slope of the regression R 3.5.1 statistical software (R Development Core Team representing how well the respective density estimates 2018), using p ≤ 0.05 to determine statistical significance. predict each other. Finally, a modified t-test was used to The fish density estimates from the three techniques were test whether the slope of the regression was statistically strongly correlated (Pearson R > 0.88; Figure 1; Table 2). different from 1, indicating whether the density estimates However, the P. afer density estimates from the video generated by the respective survey techniques had a 1:1 footage, both from the single camera footage in the deepest relationship (see Supplementary Material for R code). section of the pool and summed over the five cameras in Acclimation and camera deployment time recommended the pool, were strongly correlated with the snorkel survey by Ellender et al. (2012) were validated as follows. For the and photograph-based P. afer density estimates, but the

3.0 (a) Video vs Snorkel 3.0 (b) Photo vs Snorkel 40 (c) Photo vs Video ) ) 2 2 − − 2.5 2.5 m m s l s 30 d u a l d u a v i v i 2.0 2.0 n d i n d i i i ( (

1.5 1.5 20

1.0 1.0

PHOTO MAXN (individuals) 10

0.5 0.5 VIDEO FISH DENSITY Single camera PHOTO FISH DENSITY Five cameras 0.0 0.0 0

0.0 0.5 1.0 1.5 2.0 2.5 3.0 0.0 0.5 1.0 1.5 2.0 2.5 3.0 0 10 20 30 40 SNORKEL FISH DENSITY SNORKEL FISH DENSITY VIDEO MAXN (individuals) (individuals m−2) (individuals m−2)

Figure 1: Comparison between fish survey methods from ten pools within headwater tributaries of the Swartkops River, Eastern Cape, South Africa: a) Snorkel survey versus video fish density estimates b) Snorkel survey versus photo fish density estimates and c) Video versus photo fish abundances from five cameras in ten pools. A 1:1 relationship is indicated by the grey dotted line in each panel. The strengths of the relationships between the respective survey techniques are presented in Table 2 4 Hannweg, Marr, Bloy and Weyl

Table 2: Results of the statistical analysis to evaluate the correlation between the Pseudobarbus afer abundance estimates generated by the respective survey techniques using Pearson correlation and linear regression, including the results of t-tests to determine whether the slope of the relationship between the respective fish abundance estimates are equal to unity

Do the paired Regression Regression Regression Pearson observations come Regression slope 95% Comparison Adjusted slope correlation from the same slope confidence R2 H : β = 1 distribution? interval 0 Snorkel vs Middle Video R = 0.886, t = 6.47, df = 9, Adjusted R2 0.758; 0.485 0.278–0.692 p < 0.01 (normalised by pool area) p < 0.01 p < 0.001 p < 0.001 Snorkel vs Sum Video in pool R = 0.935, t = −2.85, df = 9, Adjusted R2 0.858 1.472 1.015–1.929 p < 0.05 (normalised by pool area) p < 0.001 p < 0.05 p << 0.001 Snorkel vs Sum Photo in pool R = 0.932, t = −0.55, df = 9, Adjusted R2 0.852; 1.197 0.817–1.577 p = 0.270 (normalised by pool area) p < 0.001 p = 0.600 p << 0.001 Video vs photo R = 0.963, t = 8.33, df = 49, Adjusted R2 0.935; 0.753 0.696–0.809 p << 0.001 p << 0.001 p << 0.001 p << 0.001

regression slopes differed significantly from 1 (Table 2, time of 3 min and the filming time of 15 min are considered Figure 1a). Video based P. afer density estimates obtained adequate for the current study system, the acclimation from a single camera set in the deepest portion of the pool time required for the study of fish assemblage should be estimated only about 50% of the standard snorkel estimate, validated at the beginning of each study. For example, whereas those summed over five microhabitats in a pool, Ebner et al. (2017) required an acclimation period of about overestimated the snorkel survey results by almost 50% 30 min, the use of multiple cameras. (Figure 1a; Table 2). The regression slope for the snorkel Pseudobarbus afer abundance was significantly higher survey and the photograph-based P. afer density estimates in the deeper and wooded habitats than in the shallower did not differ significantly from 1 (Figure 1b; Table 2), inflow and outflow habitats of the pool (ANOVA; F(4,45) = implying a 1:1 relationship between the P. afer density 6.24; p < 0.001; Figure 2; Table S2 in the Supplementary estimates of these survey techniques. Photograph and Video Material). Abundance in fern root habitat was not MaxN estimates were strongly correlated, but the regression significantly different to either the shallow or the deeper slope differed significantly from 1 with photograph estimates habitats (Figure 2; Table S2 Supplementary Material). being approximately 75% of the video MaxN estimates These data support snorkel survey observations by (Figure 1c; Table 2). Ellender et al. (2018) and minnow trap catch per unit Photograph-derived estimates of P. afer abundance effort data (Kadye and Booth 2014), which infer a diurnal summed across microhabitats not only closely reflect the preference for deeper and structural habitats by P. afer. snorkel survey estimates, they also reflect the abundance Therefore, action cameras should be considered as a of P. afer in specific microhabitats within the pool. Previous potential alternative to conventional methods for studying research on the use of UVA compared single fixed camera P. afer, e.g. snorkel surveys, electrofishing and netting. data with snorkel counts (Fulton et al. 2012); demonstrated This reinforces the idea that species-specific validations, that single fixed cameras are suboptimal for detecting such as the current study, should be conducted to confirm species and estimating their microhabitat associations the acclimation and filming times for remote camera (Ebner et al. 2015); and used stratified or habitat specific studies of stream fish; e.g. Ebner and Morgan (2013) and placement of cameras, e.g. to examine water depth effects Ebner et al. (2017). (Ebner and Morgan 2013; Cousins et al. 2017). The current study expands on these studies through the use of multiple Acknowledgements — The authors thank the staff of the Groendal cameras to generate a summed metric of density, scaled Nature Reserve for their assistance during the fieldwork. The across pools, a concept piloted by Ebner et al. (2017) for a current study was financially supported by the National Research Foundation (NRF) - South African Research Chairs Initiative of the single pool. Department of Science and Innovation (DSI) (Grant No. 110507), With the exception of one pool (Fernkloof 6, t-test; t = Water Research Commission of South Africa (K5/2538), the NRF −4.398; df = 13.91; p < 0.001), no significant differences Professional Development Programme (Grant No. 1010140), the were detected between the first 3 min post acclimation and Deutscher Akademischer Austauschdienst (DAAD) and the DSI/ the remainder of the 15-min recording time (Table S1 and NRF Centre of Excellence in Invasion Biology. We acknowledge Figure S1 in the Supplementary Material). Accumulation the funding channelled through the NRF-SAIAB Institutional curves expressing the MaxN for each 30 s time interval Support system and logistic support provided by the NRF-SAIAB as a proportion of the largest MaxN from all time intervals research platform and the SAIAB Margret Smith Library. The current (PMaxN), demonstrated that the optimal deployment study was conducted in accordance with SAIAB ethics clearance time (when PMaxN ~0.9) did not differ between camera 25.4.1.7.5_2017-04 and Eastern Cape Department of Economic Development, Environmental Affairs and Tourism permit numbers placements (t-test; t = 0.191; df = 4; p = 0.86). On average, CRO 37/17CR and CRO 38/17CR. Any opinion, finding and 0.9 MaxN was reached within 5 min, for all except the conclusion or recommendation expressed in this material is that of woody debris habitats (Figure S2 Supplementary Material). the author(s) and the NRF and Rhodes University do not accept any However, although Ellender et al. (2012)’s acclimation liability in this regard. African Journal of Aquatic Science 2020, 45(1): xxx–xxx 5

30

b

25 b

20 ER F

15 ab

MAXN OF P. A 10 a a

5

0 Inflow Woody Debris Deep Middle Fern Root Outflow

HABITAT

Figure 2: MaxN density of Pseudobarbus afer taken from underwater video analysis, grouped by habitat for ten pools in the Fernkloof and Blindekloof tributaries of the Swartkops River. Habitats that the ANOVA and post hoc Tukey HSD test identified as not being significantly different to each other are grouped by the same lower-case letter. The results of the Tukey HSD test are presented in Table S2 of the Supplementary Material

ORCIDs Cousins S, Kennard MJ, Ebner BC. 2017. Depth-related composition and structuring of tropical riverine fish assemblages revealed by SM Marr: https://orcid.org/0000-0001-8655-5522 baited video. Marine and Freshwater Research 68: 1965–1975. OLF Weyl: https://orcid.org/0000-0002-8935-3296 Cowx IG. 2002. Analysis of threats to freshwater fish conservation: Past and present challenges. (p 201–220). In: Collares-Pereira MJ, Cowx IG, Coelho MM (Eds). Conservation of freshwater References fishes: options for the future. Oxford, UK: Fishing News Books, Blackwell Science. Abell R, Thieme ML, Revenga C, Bryer M, Kottelat M, Bogutskaya Darwall WRT, Smith KG, Allen DJ, Holland RA, Harrison IJ, Brooks N, et al. 2008. Freshwater Ecoregions of the world: a new map EGE (Eds). 2011. The diversity of life in African freshwaters: under of biogeographic units for freshwater biodiversity conservation. water, under threat. An analysis of the status and distribution of Bioscience 58: 403–414. freshwater species throughout mainland Africa. Cambridge, United Anderson AA, Hubbs C, Winemiller KO, Edwards RJ. 1995. Kingdom and Gland, Switzerland: IUCN. Texas freshwater fish assemblages following three decades of Darwall WRT, Smith KG, Allen DJ, Seddon M, McGregor Reid G, environmental change. The Southwestern Naturalist 40: 314–321. Clausnitzer V, Kalkman V. 2008. Freshwater Biodiversity - A Balian EV, Segers H, Lévèque C, Martens K. 2008. The Freshwater hidden resource under threat. (p 43–54). In: Vie JC, Hinton-Taylor Diversity Assessment: an overview of the results. C, Stuart SN (Eds). Wildlife in a Changing World: An Analysis Hydrobiologia 595: 627–637. of the 2008 IUCN Red List of Threatened Species. Gland, Brydges NM, Boulcott P, Ellis T, Braithwaite VA. 2009. Quantifying Switzerland: IUCN. stress responses induced by different handling methods in three Dolloff A, Kershner J, Thurow R. 1996. Underwater observation. (p species of fish. Applied Animal Behaviour Science 116: 295–301. 533–554). In: Murphy BR, Willis DW (Eds). Fisheries Techniques. Cappo M, Harvey E, Malcom H, Speare P. 2003. Potential of video Bethesda, Maryland, USA: American Fisheries Society. techniques to monitor diversity, abundance and size of fish in Duncan JR, Lockwood JL. 2001. Extinction in the field of bullets: A studies of marine protected areas. In: Beumer JP, Grant A, Smith search for causes in the decline of the world’s freshwater fishes. DC (Eds). Aquatic Protected Areas: What works best and why? Biological Conservation 102: 97–105. Proceedings of the World Congress on Aquatic Protected Areas, Ebner BC, Clear R, Godschalx S, Beitzel M. 2009. In-stream Cairns, Australia, August 2002. Australian Society for Fish Biology, behaviour of threatened fishes and their food organisms based on Brisbane, Australia, pp 455–464. remote video monitoring. Aquatic Ecology 43: 569–576. Chakona A, Skelton PH. 2017. A review of the Pseudobarbus afer Ebner BC, Donaldson JA, Allen GR, Keith P. 2017. Visual census, (Peters, 1864) species complex (Teleostei, ) in the photographic records and the trial of a video network provide first eastern Cape Fold Ecoregion of South Africa. ZooKeys 657: evidence of the elusive Sicyopterus cynocephalus in Australia. 109–140. Cybium 41: 117–125. Chakona A, Weyl OLF, Ellender BR. 2017. Pseudobarbus afer. Ebner BC, Fulton CJ, Cousins S, Donaldson JA, Kennard MJ, The IUCN Red List of Threatened Species 2017 e.T10765 Meynecke J-O, Schaffer J. 2015. Filming and snorkelling as visual 4052A107654388. https://dx.doi.org/10.2305/IUCN.UK.2017-3. techniques to survey fauna in difficult to access tropical rainforest RLTS.T107654052A107654388.en. [Accessed 17 April 2020]. streams. Marine and Freshwater Research 66: 120–126. 6 Hannweg, Marr, Bloy and Weyl

Ebner BC, Morgan DL. 2013. Using remote underwater video to Hickey MA, Closs GP. 2006. Evaluating the potential of night estimate freshwater fish species richness. Journal of Fish Biology spotlighting as a method for assessing species composition and 82: 1592–1612. brown trout abundance: a comparison with electrofishing in small Ebner BC, Starrs D, Morgan DL, Fulton CJ, Donaldson JA, Doody streams. Journal of Fish Biology 69: 1513–1523. JS, et al. 2014. Emergence of field-based underwater video for Jordan F, Jelks HL, Bortone SA, Dorazio RM. 2008. Comparison of understanding the ecology of freshwater fishes and crustaceans visual survey and seining methods for estimating abundance of an in Australia. Journal of the Royal Society of Western Australia 97: endangered, benthic stream fish. Environmental Biology of Fishes 287–296. 81: 313–319. Ellender BR, Becker A, Weyl OLF, Swartz ER. 2012. Underwater Kadye WT, Booth AJ. 2014. Alternative responses to predation in video analysis as a non-destructive alternative to electrofishing two headwater stream minnows is reflected in their contrasting diel for sampling imperilled headwater stream fishes. Aquatic activity patterns. PLoS ONE 9(4): e93666. Conservation 22: 58–65. Limburg KE, Hughes RM, Jackson DC, Czech B. 2011. Human Ellender BR, Wasserman RJ, Chakona A, Skelton PH, Weyl OLF. population increase, economic growth, and fish conservation: 2017. A review of the biology and status of Cape Fold Ecoregion collision course or savvy stewardship? Fisheries (Bethesda, Md.) freshwater fishes. Aquatic Conservation: 27: 867–879. 36: 27–35. Ellender BR, Weyl OLF, Alexander ME, Luger AM, Nagelkerke LAJ, R Development Core Team (2018) R: A Language and Environment Woodford DJ. 2018. Out of the pot and into the fire: Explaining for Statistical Computing. Available at: http://www.R-project.org the vulnerability of an endangered small headwater stream fish to [Accessed 3 July 2018] black-bass spp. invasion. Journal of Fish Biology 92: Snyder DE. 2003. Electrofishing and its harmful effects on fish. 1035–1050. Information and Technology Report USGS/BRD/ITR-2003–0002. Ellender BR, Weyl OLF, Swartz ER. 2011. Invasion of a headwater Denver, CO, USA: US Government Printing Office. 149 p. stream by non-native fishes in the Swartkops River system, South Van Hecke T. 2012. Power study of anova versus Kruskal-Wallis Africa. African Zoology 46: 39–46. test. Journal of Statistics and Management Systems 15: 241–247. Fulton CJ, Starrs D, Ruibal MP, Ebner BC. 2012. Counting crayfish: Weyl OLF, Ellender BR, Woodford DJ, Jordaan MS. 2013. Fish active searching and baited cameras trump conventional hoop distributions in the Rondegat River, Cape Floristic Region, South netting in detecting Euastacus armatus. Endangered Species Africa, and the immediate impact of rotenone treatment in an Research 19: 39–45. invaded reach. African Journal of Aquatic Science 38: 201–209.

Manuscript received: 28 November 2018; revised: 22 November 2019 accepted: 28 November 2019 Associate Editor: J Simaika